The alcohol withdrawal syndrome (AWS) has been treated for nearly forty years with the remarkably safe and effective benzodiazepines (BZs). In the outpatient treatment of AWS, interactions with alcohol and the abuse liability of BZs invite comparisons with agents that do not share these susceptibilities. In our original Alcohol Research Center component comparing Lorazepam (LZ) to Carbamazepine (CBZ), the latter was significantly more effective in suppressing the acoustic startle response, and anxiety and depressive symptoms during treatment. In the immediate seven-day post treatment, LZ subjects showed higher levels of multiple dimensions of AWS symptoms, including higher withdrawal scores, poor sleep, and greater subjective discomfort than CBZ. CBZ, useful in research has limited clinical applications because of its widespread interactions and uncommon but serious toxicities. In the present application, we propose a double-blind controlled trial of gabapentin (GBP, 80) vs. LZ (N=80) in treatment seeking outpatients experiencing AWS. Multiple domains of AWS will be evaluated during a five-day outpatient treatment interval and a seven-day post treatment phase. Serial measures include electrophysiologic tests of eye blink acoustic startle response, aggregate withdrawal symptoms as evaluated by the CIWA-Ar, subjective assessments of sleep, global discomfort, anxiety, depression, and need for additional medications. Side effects and safety will be compared. GMP, a unique anticonvulsant, is not a controlled substance, soon will be generic, has benign side effects, has no interactions with other drugs, and has limited interactions with alcohol. Side effects and safety will be compared. GBP, a unique anticonvulsant, is not a controlled substance, soon will be generic, has benign side effects, has no interactions with other drugs, and has limited interactions with alcohol. Should GBP be superior to LZ for the treatment of AWS, it could revise the way AWS is managed in outpatients.